Treatment of acute myeloid leukemia models by targeting a cell surface RNA-binding protein.

Immunotherapies for acute myeloid leukemia (AML) and other cancers are limited by a lack of tumor-specific targets. Here we discover that RNA-binding proteins and glycosylated RNAs (glycoRNAs) form precisely organized nanodomains on cancer cell surfaces. We characterize nucleophosmin (NPM1) as an abundant cell surface protein (csNPM1) on a variety of tumor types. With a focus on AML, we observe csNPM1 on blasts and leukemic stem cells but not on normal hematopoietic stem cells. We develop a monoclonal antibody to target csNPM1, which exhibits robust anti-tumor activity in multiple syngeneic and xenograft models of AML, including patient-derived xenografts, without observable toxicity. We find that csNPM1 is expressed in a mutation-agnostic manner on primary AML cells and may therefore offer a general strategy for detecting and treating AML. Surface profiling and in vivo work also demonstrate csNPM1 as a target on solid tumors. Our data suggest that csNPM1 and its neighboring glycoRNA-cell surface RNA-binding protein (csRBP) clusters may serve as an alternative antigen class for therapeutic targeting or cell identification.